Eyeballing the components easily reveals the designers intent.
It's only intended to be able to apply significant forcing in one direction, as in moving the wheel rearward while the wheel is still clamped axially to some degree by the axle nut.
This is all evident by two thing:
There is only one thrust face on the adjuster bolt, and it only sees service when forcing the wheel rearward.
Forward forcing can only be under very light load, as it's reliant upon the circlip being a thrust surface.
In other words, only when the rear wheel is less than snug in the slots, with the axle nut backed off.
Keep in mind that the vast majority of the 919 fleet world wide, does not have centre stands nor race stands in their owners garage - IF they even have a garage.
Honda envisions a rear wheel that is weighted on a leaning bike, whose rear wheel is still aligned, that just needs some adjusting to compensate for some chain wear.
In this case, the adjuster design is A OK.
Personally, when it comes to positioning the rear wheel for proper chain adjustment, I feel that getting the wheel aligned first is the backbone of the job, done so with the chain loose, such that all subsequent adjusting movement is rearward.
Now to the circlip.
Most high volume circlips are designed to retain in one direction only.
Only one side is ground or stamped flat with no radius at the ID.
The other side will be radiused around the ID, and if that side sees thrust, the radius can too easily ride up on the groove and pop off. (Hence old school hot rod engines with floating piston pins would see back to back dual circlips at each end of the pin.)
It turns out my circlip pliers are at our son's, so I can't check mine but my assumption is that the 919's are low cost items with one side flat and one side radiused at the ID.
Make sure the flat side faces the hex headed end of the adjuster.